Predicting ship fuel consumption using artificial intelligence based on real-time data

• Main Author: Wong, Qing Er
• Other Authors: Xu Yan
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2024
• Subjects: Engineering; Ship fuel consumption; Machine learning; Artificial neural network; Prediction models
• Online Access: https://hdl.handle.net/10356/176413
• Institution: Nanyang Technological University

This paper examines the significance of accurate predictions of ship fuel consumption, highlighting its role in cost reduction and mitigating carbon dioxide emissions. While container ships have been extensively researched, there has been a noticeable lack of studies done on passenger ferries. Hence, this study conducts a comparative analysis of various Artificial Intelligence methods for predicting fuel consumption in passenger ferries. The analysis includes outlier detection using KNearest Neighbours, and employs models such as Multiple Linear Regression, Lasso Regression, XGBoost, and Artificial Neural Network in making the predictions. Performance evaluation metrics, including coefficients of determination and root mean squared error, are utilized to assess the model's performance. The findings reveal that XGBoost and Artificial Neural Network achieve the highest accuracy in predicting fuel consumption.